Prototyping

Cursor's Live Apps Solve Exposure, but Nobody Owns the App Afterward

Cursor's 5 November Build Live Apps puts internal tools on a hosted, private-by-default footing. Its docs cover publishing and inactivity, but not who owns an app once its builder leaves.

Every internal tool is software someone has to own. That is Liviu Tudor’s line , and his picture of where this leads is bleaker: apps “each built by someone who has since left, each holding some data nobody can find”. Cursor ’s 5 November launch of Build Live Apps makes that picture easier to reach. Anyone in sales, operations, legal or finance can now turn a prompt into a live app that holds real company data, but Cursor’s rules say almost nothing about who looks after it once its builder is gone.

A design that answers the leaks

Cursor’s changelog entry says “anyone in your org can now build useful tools for their work and share them with teammates”. Its examples are event sign-ups, desk seating, onboarding portals and live sales pipeline dashboards. Checks run before publishing, and “Nothing goes live until the checks pass and you explicitly publish.”

The privacy defaults look written with last spring in mind. In May, eWeek reported that Red Access found about 380,000 publicly accessible assets built with Lovable, Replit, Netlify and Base44. About 5,000 exposed potentially sensitive data. Cursor’s Apps documentation disables public sharing by default for Teams and Enterprise, and admins can see every app and secret and unpublish any app.

The pitch that “most apps and tools that users build never make it off of localhost” is also the sales case. It names a real barrier. It was also, quietly, how a company avoided owning software nobody had agreed to maintain.

Two moments governed, one left open

Read the docs closely and the lifecycle rules cluster around two moments. One is publication: checks, review rules, approvals. The other is death by neglect. On Enterprise, apps with no activity for the retention period are unpublished, the owner is emailed 30 days before, and an unpublished app’s secrets and database “can’t be recovered”.

Between those two moments sits the one where internal tools usually become a problem. Consider a mechanism, not an event. Say a sales analyst publishes a pipeline dashboard wired to the data warehouse with a team secret, then leaves. Removing a member, per Cursor’s members documentation , “removes their access to the team” but “does not delete their Cursor account or the data attached to that account”. The page does not mention Apps.

The dashboard keeps getting daily use, so the inactivity rule never fires. The Apps docs say “only the owner can change the app’s settings”, while editors “can change the app’s code, including code that uses your secrets and API keys, and publish it”. Nothing in either page says whether an admin can take over ownership. Silence is not proof it can’t be done.

The hard part of turning a prototype into a product has moved from deployment and exposure to ownership.

What this changes for the person who built it

The ops lead or sales analyst who once handed a half-built tool to engineering can now publish it themselves. That makes them the owner of live company software. Every editor they invite can change code that touches the team’s secrets.

Tudor recommends what any maintenance-minded team would: a named owner, a repo, a review step and a removal date for unused apps. He says admin controls help but governance is a habit teams must build. A DEV Community post lands in the same place: “The ownership half is now the interesting part — and it is still yours.”

Those are practitioners’ predictions and published rules. No source has observed an orphaned Cursor app. Read together, the rules suggest that the next failure mode is an app in daily use, holding live data, with no one accountable. That is a projection from documentation plus the earlier exposure wave on other platforms, not a measured outcome.

The fair reading is more flattering to Cursor

The counterargument is strong. Compared with shadow spreadsheets, scripts on laptops and public Lovable or Replit apps, Cursor’s setup is arguably far better governed. Admins see everything, team apps sit behind company sign-in, and the inactivity rule is itself a lifecycle policy. Cursor may also handle ownership transfer through admins or support without documenting it.

Replit’s CEO Amjad Masad argues the opposite philosophy. In a statement cited by WIRED and reported by eWeek , he said “Public apps being accessible on the internet is expected behavior”, which makes configuration the user’s job. Cursor plainly disagrees, and its defaults are better for it.

The hard part of turning a prototype into a product has moved from deployment and exposure to ownership. Better defaults fix exposure, not custody. A prototype on a laptop dies when its author stops caring. A hosted app with a login and a database keeps running. When deployment costs a prompt, the badge on the builder’s lanyard stops being the only thing tying them to the tool, and someone has to decide who inherits the binder.

This article was written by AI. How Pipeline works.